def _unjsonify(x, isattributes=False):
"""Convert JSON string to an ordered defaultdict."""
if isattributes:
obj = simplejson.loads(x)
return feature.dict_class(obj)
return simplejson.loads(x)
def _split_keyvals(keyval_str, dialect=None):
"""
Given the string attributes field of a GFF-like line, split it into an
attributes dictionary and a "dialect" dictionary which contains information
needed to reconstruct the original string.
Lots of logic here to handle all the corner cases.
If `dialect` is None, then do all the logic to infer a dialect from this
attribute string.
Otherwise, use the provided dialect (and return it at the end).
"""
infer_dialect = False
if dialect is None:
# Make a copy of default dialect so it can be modified as needed
dialect = copy.copy(constants.dialect)
infer_dialect = True
quals = feature.dict_class()
if not keyval_str:
return quals, dialect
# If a dialect was provided, then use that directly.
if not infer_dialect:
if dialect["trailing semicolon"]:
keyval_str = keyval_str.rstrip(";")
parts = keyval_str.split(dialect["field separator"])
kvsep = dialect["keyval separator"]
if dialect["leading semicolon"]:
pieces = []
for p in parts:
if p and p[0] == ";":
p = p[1:]
pieces.append(p.strip().split(kvsep))
key_vals = [(p[0], " ".join(p[1:])) for p in pieces]
if dialect["fmt"] == "gff3":
key_vals = [p.split(kvsep) for p in parts]
else:
leadingsemicolon = dialect["leading semicolon"]
pieces = []
for i, p in enumerate(parts):
if i == 0 and leadingsemicolon:
p = p[1:]
pieces.append(p.strip().split(kvsep))
key_vals = [(p[0], " ".join(p[1:])) for p in pieces]
quoted = dialect["quoted GFF2 values"]
for item in key_vals:
# Easy if it follows spec
if len(item) == 2:
key, val = item
# Only key provided?
else:
assert len(item) == 1, item
key = item[0]
val = ""
try:
quals[key]
except KeyError:
quals[key] = []
if quoted:
if len(val) > 0 and val[0] == '"' and val[-1] == '"':
val = val[1:-1]
if val:
# TODO: if there are extra commas for a value, just use empty
# strings
# quals[key].extend([v for v in val.split(',') if v])
vals = val.split(",")
quals[key].extend(vals)
return quals, dialect
# If we got here, then we need to infer the dialect....
#
# Reset the order to an empty list so that it will only be populated with
# keys that are found in the file.
dialect["order"] = []
# ensembl GTF has trailing semicolon
if keyval_str[-1] == ";":
keyval_str = keyval_str[:-1]
dialect["trailing semicolon"] = True
# GFF2/GTF has a semicolon with at least one space after it.
# Spaces can be on both sides (e.g. wormbase)
# GFF3 works with no spaces.
# So split on the first one we can recognize...
for sep in (" ; ", "; ", ";"):
parts = keyval_str.split(sep)
if len(parts) > 1:
dialect["field separator"] = sep
break
# Is it GFF3? They have key-vals separated by "="
if gff3_kw_pat.match(parts[0]):
key_vals = [p.split("=") for p in parts]
dialect["fmt"] = "gff3"
dialect["keyval separator"] = "="
# Otherwise, key-vals separated by space. Key is first item.
else:
dialect["keyval separator"] = " "
pieces = []
for p in parts:
# Fix misplaced semicolons in keys in some GFF2 files
if p and p[0] == ";":
p = p[1:]
dialect["leading semicolon"] = True
pieces.append(p.strip().split(" "))
key_vals = [(p[0], " ".join(p[1:])) for p in pieces]
for item in key_vals:
# Easy if it follows spec
if len(item) == 2:
key, val = item
# Only key provided?
else:
assert len(item) == 1, item
key = item[0]
val = ""
# Is the key already in there?
if key in quals:
dialect["repeated keys"] = True
else:
quals[key] = []
# Remove quotes in GFF2
if len(val) > 0 and val[0] == '"' and val[-1] == '"':
val = val[1:-1]
dialect["quoted GFF2 values"] = True
if val:
# TODO: if there are extra commas for a value, just use empty
# strings
# quals[key].extend([v for v in val.split(',') if v])
vals = val.split(",")
if (len(vals) > 1) and dialect["repeated keys"]:
raise helpers.AttributeStringError(
"Internally inconsistent attributes formatting: " "some have repeated keys, some do not."
)
quals[key].extend(vals)
# keep track of the order of keys
dialect["order"].append(key)
# for key, vals in quals.items():
#
# TODO: urllib.unquote breaks round trip invariance for "hybrid1.gff3"
# test file. This is because the "Note" field has %xx escape chars,
# but "Dbxref" has ":" which, if everything were consistent, should
# have also been escaped.
#
# (By the way, GFF3 spec says only literal use of \t, \n, \r, %, and
# control characters should be encoded)
#
# Solution 1: don't unquote
# Solution 2: store, along with each attribute, whether or not it
# should be quoted later upon reconstruction
# Solution 3: don't care about invariance
# unquoted = [urllib.unquote(v) for v in vals]
# quals[key] = vals
if (dialect["keyval separator"] == " ") and (dialect["quoted GFF2 values"]):
dialect["fmt"] = "gtf"
return quals, dialect
