def test_unjsonify():
attributes, dialect = parser._split_keyvals('transcript_id "mRNA1"')
assert attributes == {'transcript_id': ['mRNA1']}, attributes
s = helpers._jsonify(attributes)
assert s == '{"transcript_id":["mRNA1"]}', s
d = helpers._unjsonify(s, isattributes=True)
assert d == attributes
def test_unjsonify():
attributes, dialect = parser._split_keyvals('transcript_id "mRNA1"')
assert attributes == {'transcript_id': ['mRNA1']}, attributes
s = helpers._jsonify(attributes)
assert s == '{"transcript_id":["mRNA1"]}', s
d = helpers._unjsonify(s, isattributes=True)
assert d == attributes
def feature_from_line(line, dialect=None, strict=True, keep_order=False):
"""
Given a line from a GFF file, return a Feature object
Parameters
----------
line : string
strict : bool
If True (default), assume `line` is a single, tab-delimited string that
has at least 9 fields.
If False, then the input can have a more flexible format, useful for
creating single ad hoc features or for writing tests. In this case,
`line` can be a multi-line string (as long as it has a single non-empty
line), and, as long as there are only 9 fields (standard GFF/GTF), then
it's OK to use spaces instead of tabs to separate fields in `line`.
But if >9 fields are to be used, then tabs must be used.
keep_order, dialect
Passed directly to :class:`Feature`; see docstring for that class for
description
Returns
-------
A new :class:`Feature` object.
"""
if not strict:
lines = line.splitlines(False)
_lines = []
for i in lines:
i = i.strip()
if len(i) > 0:
_lines.append(i)
assert len(_lines) == 1, _lines
line = _lines[0]
if '\t' in line:
fields = line.rstrip('\n\r').split('\t')
else:
fields = line.rstrip('\n\r').split(None, 8)
else:
fields = line.rstrip('\n\r').split('\t')
try:
attr_string = fields[8]
except IndexError:
attr_string = ""
attrs, _dialect = parser._split_keyvals(attr_string, dialect=dialect)
d = dict(list(zip(constants._gffkeys, fields)))
d['attributes'] = attrs
d['extra'] = fields[9:]
d['keep_order'] = keep_order
if dialect is None:
dialect = _dialect
return Feature(dialect=dialect, **d)
def feature_from_line(line, dialect=None, strict=True, keep_order=False):
"""
Given a line from a GFF file, return a Feature object
Parameters
----------
line : string
strict : bool
If True (default), assume `line` is a single, tab-delimited string that
has at least 9 fields.
If False, then the input can have a more flexible format, useful for
creating single ad hoc features or for writing tests. In this case,
`line` can be a multi-line string (as long as it has a single non-empty
line), and, as long as there are only 9 fields (standard GFF/GTF), then
it's OK to use spaces instead of tabs to separate fields in `line`.
But if >9 fields are to be used, then tabs must be used.
keep_order, dialect
Passed directly to :class:`Feature`; see docstring for that class for
description
Returns
-------
A new :class:`Feature` object.
"""
if not strict:
lines = line.splitlines(False)
_lines = []
for i in lines:
i = i.strip()
if len(i) > 0:
_lines.append(i)
assert len(_lines) == 1, _lines
line = _lines[0]
if '\t' in line:
fields = line.rstrip('\n\r').split('\t')
else:
fields = line.rstrip('\n\r').split(None, 8)
else:
fields = line.rstrip('\n\r').split('\t')
try:
attr_string = fields[8]
except IndexError:
attr_string = ""
attrs, _dialect = parser._split_keyvals(attr_string, dialect=dialect)
d = dict(list(zip(constants._gffkeys, fields)))
d['attributes'] = attrs
d['extra'] = fields[9:]
d['keep_order'] = keep_order
if dialect is None:
dialect = _dialect
return Feature(dialect=dialect, **d)
def infer_dialect(attributes):
"""
Infer the dialect based on the attributes.
Parameters
----------
attributes : str or iterable
A single attributes string from a GTF or GFF line, or an iterable of
such strings.
"""
if isinstance(attributes, six.string_types):
attributes = [attributes]
dialects = [parser._split_keyvals(i)[1] for i in attributes]
return _choose_dialect(dialects)
def infer_dialect(attributes):
"""
Infer the dialect based on the attributes.
Parameters
----------
attributes : str or iterable
A single attributes string from a GTF or GFF line, or an iterable of
such strings.
"""
if isinstance(attributes, six.string_types):
attributes = [attributes]
dialects = [parser._split_keyvals(i)[1] for i in attributes]
return _choose_dialect(dialects)
def attrs_OK(attr_str, attr_dict, acceptable_reconstruction=None):
"""
Given an attribute string and a dictionary of what you expect, test the
attribute splitting and reconstruction (invariant roundtrip).
There are some corner cases for the roundtrip invariance that don't work
(see attr_test_cases.py for details); `acceptable_reconstruction` handles
those.
"""
result, dialect = parser._split_keyvals(attr_str)
assert result == attr_dict, result
reconstructed = parser._reconstruct(result, dialect, keep_order=True)
if acceptable_reconstruction:
assert reconstructed == acceptable_reconstruction, reconstructed
else:
assert reconstructed == attr_str, reconstructed
def attrs_OK(attr_str, attr_dict, acceptable_reconstruction=None):
"""
Given an attribute string and a dictionary of what you expect, test the
attribute splitting and reconstruction (invariant roundtrip).
There are some corner cases for the roundtrip invariance that don't work
(see attr_test_cases.py for details); `acceptable_reconstruction` handles
those.
"""
result, dialect = parser._split_keyvals(attr_str)
assert result == attr_dict, result
reconstructed = parser._reconstruct(result, dialect, keep_order=True)
if acceptable_reconstruction:
assert reconstructed == acceptable_reconstruction, reconstructed
else:
assert reconstructed == attr_str, reconstructed
def __init__(self, seqid=".", source=".", featuretype=".",
start=".", end=".", score=".", strand=".", frame=".",
attributes=None, extra=None, bin=None, id=None, dialect=None,
file_order=None, keep_order=False,
sort_attribute_values=False):
"""
Represents a feature from the database.
Usually you won't want to use this directly, since it has various
implementation details needed for operating in the context of FeatureDB
objects. Instead, try the :func:`gffutils.feature.feature_from_line`
function.
When printed, reproduces the original line from the file as faithfully
as possible using `dialect`.
Parameters
----------
seqid : string
Name of the sequence (often chromosome)
source : string
Source of the feature; typically the originating database or
program that predicted the feature
featuretype : string
Type of feature. For example "gene", "exon", "TSS", etc
start, end : int or "."
1-based coordinates; start must be <= end. If "." (the default
placeholder for GFF files), then the corresponding attribute will
be None.
score : string
Stored as a string.
strand : "+" | "-" | "."
Strand of the feature; "." when strand is not relevant.
frame : "0" | "1" | "2"
Coding frame. 0 means in-frame; 1 means there is one extra base at
the beginning, so the first codon starts at the second base;
2 means two extra bases at the beginning. Interpretation is strand
specific; "beginning" for a minus-strand feature is at the end
coordinate.
attributes : string or dict
If a string, first assume it is serialized JSON; if this fails then
assume it's the original key/vals string. If it's a dictionary
already, then use as-is.
The end result is that this instance's `attributes` attribute will
always be a dictionary.
Upon printing, the attributes will be reconstructed based on this
dictionary and the dialect -- except if the original attributes
string was provided, in which case that will be used directly.
extra : string or list
Additional fields after the canonical 9 fields for GFF/GTF.
If a string, then first assume it's serialized JSON; if this fails
then assume it's a tab-delimited string of additional fields. If
it's a list already, then use as-is.
bin : int
UCSC genomic bin. If None, will be created based on provided
start/end; if start or end is "." then bin will be None.
id : None or string
Database-specific primary key for this feature. The only time this
should not be None is if this feature is coming from a database, in
which case it will be filled in automatically.
dialect : dict or None
The dialect to use when reconstructing attribute strings; defaults
to the GFF3 spec. :class:`FeatureDB` objects will automatically
attach the dialect from the original file.
file_order : int
This is the `rowid` special field used in a sqlite3 database; this
is provided by FeatureDB.
keep_order : bool
If True, then the attributes in the printed string will be in the
order specified in the dialect. Disabled by default, since this
sorting step is time-consuming over many features.
sort_attribute_values : bool
If True, then the values of each attribute will be sorted when the
feature is printed. Mostly useful for testing, where the order is
important for checking against expected values. Disabled by
default, since it can be time-consuming over many features.
"""
# start/end can be provided as int-like, ".", or None, but will be
# converted to int or None
if start == ".":
start = None
elif start is not None:
start = int(start)
if end == ".":
end = None
elif end is not None:
end = int(end)
# Flexible handling of attributes:
# If dict, then use that; otherwise assume JSON and convert to a dict;
# otherwise assume original string and convert to a dict.
#
# dict_class is set at the module level above...this is so you can swap
# in and out different dict implementations (ordered, defaultdict, etc)
# for testing.
attributes = attributes or dict_class()
if isinstance(attributes, six.string_types):
try:
attributes = helpers._unjsonify(attributes, isattributes=True)
# it's a string but not JSON: assume original attributes string.
except simplejson.JSONDecodeError:
# But Feature.attributes is still a dict
attributes, _dialect = parser._split_keyvals(attributes)
# Use this dialect if none provided.
dialect = dialect or _dialect
# If string, then try un-JSONifying it into a list; if that doesn't
# work then assume it's tab-delimited and convert to a list.
extra = extra or []
if isinstance(extra, six.string_types):
try:
extra = helpers._unjsonify(extra)
except simplejson.JSONDecodeError:
extra = extra.split('\t')
self.seqid = seqid
self.source = source
self.featuretype = featuretype
self.start = start
self.end = end
self.score = score
self.strand = strand
self.frame = frame
self.attributes = attributes
self.extra = extra
self.bin = self.calc_bin(bin)
self.id = id
self.dialect = dialect or constants.dialect
self.file_order = file_order
self.keep_order = keep_order
self.sort_attribute_values = sort_attribute_values
def __init__(self, seqid=".", source=".", featuretype=".",
start=".", end=".", score=".", strand=".", frame=".",
attributes=None, extra=None, bin=None, id=None, dialect=None,
file_order=None, keep_order=False,
sort_attribute_values=False):
"""
Represents a feature from the database.
Usually you won't want to use this directly, since it has various
implementation details needed for operating in the context of FeatureDB
objects. Instead, try the :func:`gffutils.feature.feature_from_line`
function.
When printed, reproduces the original line from the file as faithfully
as possible using `dialect`.
Parameters
----------
seqid : string
Name of the sequence (often chromosome)
source : string
Source of the feature; typically the originating database or
program that predicted the feature
featuretype : string
Type of feature. For example "gene", "exon", "TSS", etc
start, end : int or "."
1-based coordinates; start must be <= end. If "." (the default
placeholder for GFF files), then the corresponding attribute will
be None.
score : string
Stored as a string.
strand : "+" | "-" | "."
Strand of the feature; "." when strand is not relevant.
frame : "0" | "1" | "2"
Coding frame. 0 means in-frame; 1 means there is one extra base at
the beginning, so the first codon starts at the second base;
2 means two extra bases at the beginning. Interpretation is strand
specific; "beginning" for a minus-strand feature is at the end
coordinate.
attributes : string or dict
If a string, first assume it is serialized JSON; if this fails then
assume it's the original key/vals string. If it's a dictionary
already, then use as-is.
The end result is that this instance's `attributes` attribute will
always be a dictionary.
Upon printing, the attributes will be reconstructed based on this
dictionary and the dialect -- except if the original attributes
string was provided, in which case that will be used directly.
extra : string or list
Additional fields after the canonical 9 fields for GFF/GTF.
If a string, then first assume it's serialized JSON; if this fails
then assume it's a tab-delimited string of additional fields. If
it's a list already, then use as-is.
bin : int
UCSC genomic bin. If None, will be created based on provided
start/end; if start or end is "." then bin will be None.
id : None or string
Database-specific primary key for this feature. The only time this
should not be None is if this feature is coming from a database, in
which case it will be filled in automatically.
dialect : dict or None
The dialect to use when reconstructing attribute strings; defaults
to the GFF3 spec. :class:`FeatureDB` objects will automatically
attach the dialect from the original file.
file_order : int
This is the `rowid` special field used in a sqlite3 database; this
is provided by FeatureDB.
keep_order : bool
If True, then the attributes in the printed string will be in the
order specified in the dialect. Disabled by default, since this
sorting step is time-consuming over many features.
sort_attribute_values : bool
If True, then the values of each attribute will be sorted when the
feature is printed. Mostly useful for testing, where the order is
important for checking against expected values. Disabled by
default, since it can be time-consuming over many features.
"""
# start/end can be provided as int-like, ".", or None, but will be
# converted to int or None
if start == ".":
start = None
elif start is not None:
start = int(start)
if end == ".":
end = None
elif end is not None:
end = int(end)
# Flexible handling of attributes:
# If dict, then use that; otherwise assume JSON and convert to a dict;
# otherwise assume original string and convert to a dict.
#
# dict_class is set at the module level above...this is so you can swap
# in and out different dict implementations (ordered, defaultdict, etc)
# for testing.
attributes = attributes or dict_class()
if isinstance(attributes, six.string_types):
try:
attributes = helpers._unjsonify(attributes, isattributes=True)
# it's a string but not JSON: assume original attributes string.
except simplejson.JSONDecodeError:
# But Feature.attributes is still a dict
attributes, _dialect = parser._split_keyvals(attributes)
# Use this dialect if none provided.
dialect = dialect or _dialect
# If string, then try un-JSONifying it into a list; if that doesn't
# work then assume it's tab-delimited and convert to a list.
extra = extra or []
if isinstance(extra, six.string_types):
try:
extra = helpers._unjsonify(extra)
except simplejson.JSONDecodeError:
extra = extra.split('\t')
self.seqid = seqid
self.source = source
self.featuretype = featuretype
self.start = start
self.end = end
self.score = score
self.strand = strand
self.frame = frame
self.attributes = attributes
self.extra = extra
self.bin = self.calc_bin(bin)
self.id = id
self.dialect = dialect or constants.dialect
self.file_order = file_order
self.keep_order = keep_order
self.sort_attribute_values = sort_attribute_values
def test_empty_split_keyvals():
attrs, dialect = parser._split_keyvals(keyval_str=None)
assert attrs == feature.dict_class()
assert dialect == constants.dialect
def test_empty_split_keyvals():
attrs, dialect = parser._split_keyvals(keyval_str=None)
assert attrs == feature.dict_class()
assert dialect == constants.dialect
