def test_realpath(self):
"""Test real_upath and real_ppath."""
print real_upath("ref with space")
self.assertTrue(real_upath("ref with space").endswith("ref\ with\ space"))
self.assertTrue(real_upath("ref\ with\ space").endswith("ref\ with\ space"))
self.assertTrue(real_ppath("ref with space").endswith("ref with space"))
self.assertTrue(real_ppath("ref\ with\ space").endswith("ref with space"))
def test_realpath(self):
"""Test real_upath and real_ppath."""
print real_upath("ref with space")
self.assertTrue(
real_upath("ref with space").endswith("ref\ with\ space"))
self.assertTrue(
real_upath("ref\ with\ space").endswith("ref\ with\ space"))
self.assertTrue(
real_ppath("ref with space").endswith("ref with space"))
self.assertTrue(
real_ppath("ref\ with\ space").endswith("ref with space"))
def _resolveAlgorithmOptions(self, options, fileNames):
""" Resolve options specified within --algorithmOptions with
options parsed from the command-line or the config file.
Return updated options.
If find conflicting values of the following options, error out.
(1) --maxHits and blasr -bestn
(2) --maxAnchorSize and blasr -minMatch
(3) --useccs and blasr -useccs/-useccsall/useccsdenovo
If find conflicting values of sawriter, regionTable and nproc,
it does not matter which value is used.
Input:
options : the original pbalign options from argumentList
and configFile.
fileNames: an PBAlignFiles object
Output:
new options by resolving options specified within
--algorithmOptions and the original pbalign options
"""
if options.algorithmOptions is None:
return options
ignoredBinaryOptions = ['-m', '-out', '-V']
ignoredUnitaryOptions = [
'-h', '--help', '--version', '-v', '-vv', '--sam', '--bam'
]
items = self.__parseAlgorithmOptionItems(options.algorithmOptions)
i = 0
try:
while i < len(items):
infoMsg, errMsg, item = "", "", items[i]
if item == "--sa":
val = real_upath(items[i + 1])
if fileNames.sawriterFileName != val:
infoMsg = "Over write sa file with {0}".format(val)
fileNames.sawriterFileName = val
elif item == "--regionTable":
val = real_upath(items[i + 1])
if fileNames.regionTable != val:
infoMsg = "Over write regionTable with {0}.\n"\
.format(val)
fileNames.regionTable = val
elif item == "--bestn":
val = int(items[i + 1])
if options.maxHits is not None and \
int(options.maxHits) != val:
errMsg = "blasr --bestn specified within " + \
"--algorithmOptions is equivalent to " + \
"--maxHits. Conflicting values of " + \
"--algorithmOptions '--bestn' and " +\
"--maxHits have been found."
else:
options.maxHits = val
elif item == "--minMatch":
val = int(items[i + 1])
if options.minAnchorSize is not None and \
int(options.minAnchorSize) != val:
errMsg = "blasr --minMatch specified within " + \
"--algorithmOptions is equivalent to " + \
"--minAnchorSize. Conflicting values " + \
"of --algorithmOptions '--minMatch' and " + \
"--minAnchorSize have been found."
else:
options.minAnchorSize = val
elif item == "--nproc":
val = int(items[i + 1])
# The number of threads is not critical.
if options.nproc is None or \
int(options.nproc) != val:
infoMsg = "Over write nproc with {n}.".format(n=val)
options.nproc = val
elif item == "--noSplitSubreads":
if not options.noSplitSubreads:
infoMsg = "Over write noSplitSubreads with True."
logging.info(self.name +
": Resolve algorithmOptions. " + infoMsg)
options.noSplitSubreads = True
del items[i]
continue
elif item == "--concordant":
if not options.concordant:
infoMsg = "Over writer concordant with True."
logging.info(self.name +
": Resolve algorithmOptions. " + infoMsg)
options.concordant = True
del items[i]
elif "--useccs" in item: # -useccs, -useccsall, -useccsdenovo
val = item.lstrip('--')
if options.useccs != val and options.useccs is not None:
errMsg = "Found conflicting options in " + \
"--algorithmOptions '{v}' \nand --useccs={u}"\
.format(v=item, u=options.useccs)
else:
options.useccs = val
elif item == "--unaligned":
val = str(items[i + 1])
options.unaligned = val
elif item == "--seed" or item == "--randomSeed":
val = int(items[i + 1])
if options.seed is None or int(options.seed) != val:
infoMsg = "Overwrite random seed with {0}.".format(val)
options.seed = val
elif item in ignoredBinaryOptions:
pass
elif item in ignoredUnitaryOptions:
del items[i:i + 1]
continue
else:
i += 1
continue
if errMsg != "":
logging.error(errMsg)
raise ValueError(errMsg)
if infoMsg != "":
logging.info(self.name + ": Resolve algorithmOptions. " +
infoMsg)
del items[i:i + 2]
except Exception as e:
errMsg = "An error occured during parsing algorithmOptions " + \
"'{ao}': ".format(ao=options.algorithmOptions)
logging.error(errMsg + str(e))
raise ValueError(errMsg + str(e))
# Existing suffix array always uses match size 8.
# When BLASR search option -minMatch is less than 8, suffix array needs
# to be created on the fly.
if (options.minAnchorSize is not None and options.minAnchorSize != ""
and int(options.minAnchorSize) < 8):
logging.warning("Suffix array must be recreated on the fly when " +
"minMatch < 8, which may take a long time.")
fileNames.sawriterFileName = None
# Update algorithmOptions when resolve is done
options.algorithmOptions = " ".join(items)
return options
def _resolveAlgorithmOptions(self, options, fileNames):
""" Resolve options specified within --algorithmOptions with
options parsed from the command-line or the config file.
Return updated options.
If find conflicting values of the following options, error out.
(1) --maxHits and blasr -bestn
(2) --maxAnchorSize and blasr -minMatch
(3) --useccs and blasr -useccs/-useccsall/useccsdenovo
If find conflicting values of sawriter, regionTable and nproc,
it does not matter which value is used.
Input:
options : the original pbalign options from argumentList
and configFile.
fileNames: an PBAlignFiles object
Output:
new options by resolving options specified within
--algorithmOptions and the original pbalign options
"""
if options.algorithmOptions is None:
return options
ignoredBinaryOptions = ["-m", "-out", "-V"]
ignoredUnitaryOptions = ["-h", "--help", "--version", "-v", "-vv", "-sam"]
items = self.__parseAlgorithmOptionItems(options.algorithmOptions)
i = 0
try:
while i < len(items):
infoMsg, errMsg, item = "", "", items[i]
if item == "-sa":
val = real_upath(items[i + 1])
if fileNames.sawriterFileName != val:
infoMsg = "Over write sa file with {0}".format(val)
fileNames.sawriterFileName = val
elif item == "-regionTable":
val = real_upath(items[i + 1])
if fileNames.regionTable != val:
infoMsg = "Over write regionTable with {0}.\n".format(val)
fileNames.regionTable = val
elif item == "-bestn":
val = int(items[i + 1])
if options.maxHits is not None and int(options.maxHits) != val:
errMsg = (
"blasr -bestn specified within "
+ "--algorithmOptions is equivalent to "
+ "--maxHits. Conflicting values of "
+ "--algorithmOptions '-bestn' and "
+ "--maxHits have been found."
)
else:
options.maxHits = val
elif item == "-minMatch":
val = int(items[i + 1])
if options.minAnchorSize is not None and int(options.minAnchorSize) != val:
errMsg = (
"blasr -minMatch specified within "
+ "--algorithmOptions is equivalent to "
+ "--minAnchorSize. Conflicting values "
+ "of --algorithmOptions '-minMatch' and "
+ "--minAnchorSize have been found."
)
else:
options.minAnchorSize = val
elif item == "-nproc":
val = int(items[i + 1])
# The number of threads is not critical.
if options.nproc is None or int(options.nproc) != val:
infoMsg = "Over write nproc with {n}.".format(n=val)
options.nproc = val
elif item == "-noSplitSubreads":
if not options.noSplitSubreads:
infoMsg = "Over write noSplitSubreads with True."
logging.info(self.name + ": Resolve algorithmOptions. " + infoMsg)
options.noSplitSubreads = True
del items[i]
continue
elif item == "-concordant":
if not options.concordant:
infoMsg = "Over writer concordant with True."
logging.info(self.name + ": Resolve algorithmOptions. " + infoMsg)
options.concordant = True
del items[i]
elif "-useccs" in item: # -useccs, -useccsall, -useccsdenovo
val = item.lstrip("-")
if options.useccs != val and options.useccs is not None:
errMsg = "Found conflicting options in " + "--algorithmOptions '{v}' \nand --useccs={u}".format(
v=item, u=options.useccs
)
else:
options.useccs = val
elif item == "-seed" or item == "-randomSeed":
val = int(items[i + 1])
if options.seed is None or int(options.seed) != val:
infoMsg = "Overwrite random seed with {0}.".format(val)
options.seed = val
elif item in ignoredBinaryOptions:
pass
elif item in ignoredUnitaryOptions:
del items[i : i + 1]
continue
else:
i += 1
continue
if errMsg != "":
logging.error(errMsg)
raise ValueError(errMsg)
if infoMsg != "":
logging.info(self.name + ": Resolve algorithmOptions. " + infoMsg)
del items[i : i + 2]
except Exception as e:
errMsg = "An error occured during parsing algorithmOptions " + "'{ao}': ".format(
ao=options.algorithmOptions
)
logging.error(errMsg + str(e))
raise ValueError(errMsg + str(e))
# Update algorithmOptions when resolve is done
options.algorithmOptions = " ".join(items)
return options
def _resolveAlgorithmOptions(self, options, fileNames):
""" Resolve options specified within --algorithmOptions with
options parsed from the command-line or the config file.
Return updated options.
If find conflicting values of the following options, error out.
(1) --maxHits and blasr -bestn
(2) --maxAnchorSize and blasr -minMatch
(3) --useccs and blasr -useccs/-useccsall/useccsdenovo
If find conflicting values of sawriter, regionTable and nproc,
it does not matter which value is used.
Input:
options : the original pbalign options from argumentList
and configFile.
fileNames: an PBAlignFiles object
Output:
new options by resolving options specified within
--algorithmOptions and the original pbalign options
"""
if options.algorithmOptions is None:
return options
ignoredBinaryOptions = ['-m', '-out', '-V']
ignoredUnitaryOptions = ['-h', '--help', '--version',
'-v', '-vv', '--sam', '--bam']
items = self.__parseAlgorithmOptionItems(options.algorithmOptions)
i = 0
try:
while i < len(items):
infoMsg, errMsg, item = "", "", items[i]
if item == "--sa":
val = real_upath(items[i+1])
if fileNames.sawriterFileName != val:
infoMsg = "Over write sa file with {0}".format(val)
fileNames.sawriterFileName = val
elif item == "--regionTable":
val = real_upath(items[i+1])
if fileNames.regionTable != val:
infoMsg = "Over write regionTable with {0}.\n"\
.format(val)
fileNames.regionTable = val
elif item == "--bestn":
val = int(items[i+1])
if options.maxHits is not None and \
int(options.maxHits) != val:
errMsg = "blasr --bestn specified within " + \
"--algorithmOptions is equivalent to " + \
"--maxHits. Conflicting values of " + \
"--algorithmOptions '--bestn' and " +\
"--maxHits have been found."
else:
options.maxHits = val
elif item == "--minMatch":
val = int(items[i+1])
if options.minAnchorSize is not None and \
int(options.minAnchorSize) != val:
errMsg = "blasr --minMatch specified within " + \
"--algorithmOptions is equivalent to " + \
"--minAnchorSize. Conflicting values " + \
"of --algorithmOptions '--minMatch' and " + \
"--minAnchorSize have been found."
else:
options.minAnchorSize = val
elif item == "--maxMatch":
val = int(items[i+1])
if options.maxMatch is not None and \
int(options.maxMatch) != val:
infoMsg = "Override maxMatch with {n}.".format(n=val)
options.maxMatch = val
elif item == "--nproc":
val = int(items[i+1])
# The number of threads is not critical.
if options.nproc is None or \
int(options.nproc) != val:
infoMsg = "Over write nproc with {n}.".format(n=val)
options.nproc = val
elif item == "--noSplitSubreads":
if not options.noSplitSubreads:
infoMsg = "Over write noSplitSubreads with True."
logging.info(self.name +
": Resolve algorithmOptions. " + infoMsg)
options.noSplitSubreads = True
del items[i]
continue
elif item == "--concordant":
if not options.concordant:
infoMsg = "Over writer concordant with True."
logging.info(self.name +
": Resolve algorithmOptions. " + infoMsg)
options.concordant = True
del items[i]
elif "--useccs" in item: # -useccs, -useccsall, -useccsdenovo
val = item.lstrip('--')
if options.useccs != val and options.useccs is not None:
errMsg = "Found conflicting options in " + \
"--algorithmOptions '{v}' \nand --useccs={u}"\
.format(v=item, u=options.useccs)
else:
options.useccs = val
elif item == "--unaligned":
val = str(items[i+1])
options.unaligned = val
elif item == "--seed" or item == "--randomSeed":
val = int(items[i+1])
if options.seed is None or int(options.seed) != val:
infoMsg = "Overwrite random seed with {0}.".format(val)
options.seed = val
elif item in ignoredBinaryOptions:
pass
elif item in ignoredUnitaryOptions:
del items[i:i+1]
continue
else:
i += 1
continue
if errMsg != "":
logging.error(errMsg)
raise ValueError(errMsg)
if infoMsg != "":
logging.info(self.name + ": Resolve algorithmOptions. " +
infoMsg)
del items[i:i+2]
except Exception as e:
errMsg = "An error occured during parsing algorithmOptions " + \
"'{ao}': ".format(ao=options.algorithmOptions)
logging.error(errMsg + str(e))
raise ValueError(errMsg + str(e))
# Existing suffix array always uses match size 8.
# When BLASR search option -minMatch is less than 8, suffix array needs
# to be created on the fly.
if (options.minAnchorSize is not None and options.minAnchorSize != "" and
int(options.minAnchorSize) < 8):
logging.warning("Suffix array must be recreated on the fly when " +
"minMatch < 8, which may take a long time.")
fileNames.sawriterFileName = None
# Update algorithmOptions when resolve is done
options.algorithmOptions = " ".join(items)
return options