def getTitleAndSuffixWithCompressionSuffixesRemoved(gSuiteTrack):
gSuiteReq = GSuiteRequirements(allowedLocations=[LOCAL, REMOTE],
allowedFileFormats=[PREPROCESSED, PRIMARY, UNKNOWN])
gSuiteReq.check(gSuiteTrack)
title, suffix, path = gSuiteTrack.title, gSuiteTrack.suffix, gSuiteTrack.path
if suffix:
for compSuffix in COMPRESSION_SUFFIXES:
reduceLen = len(compSuffix)+1
if suffix.lower() == compSuffix:
if path.endswith('.' + compSuffix):
path = path[:-reduceLen]
suffix = getFileSuffix(path)
else:
if isinstance(gSuiteTrack, SearchQueryForSuffixGSuiteTrack):
tempGSuiteTrack = GSuiteTrack(gSuiteTrack.uri.replace('.' + compSuffix, ''), title='')
suffix = tempGSuiteTrack.suffix
else:
suffix = None # Impossible to find uncompressed suffix
elif suffix.lower().endswith('.' + compSuffix):
suffix = suffix[:-reduceLen] # e.g. suffix = 'bed.gz' -> 'bed'
else:
continue
title = getTitleWithSuffixReplaced(title, '')
break
return title, suffix
def suffix(self):
suffix = super(SearchQueryForSuffixGSuiteTrack, self).suffix
if suffix is None:
fileName = self._getFileNameFromQuery()
if fileName:
suffix = getFileSuffix(fileName)
return suffix
def _getFileNameFromQuery(self):
import re
query = self._parsedUri.query
for part in re.split('[&;]', query):
subparts = part.split('=')
if len(subparts) == 2:
key, val = subparts
partSuffix = getFileSuffix(val)
if partSuffix != '': # Looks like a file name with suffix
fileName = val
return fileName
def suffix(self):
if self._parsedUri.params != '':
return self._parsedUri.params
else:
if self.path is None:
return None
suffix = getFileSuffix(self.path)
if suffix == '':
return None
return suffix
def renameBaseFileNameWithDuplicateIdx(baseFileName, duplicateIdx):
if duplicateIdx == 1:
return baseFileName
else:
suffixes = []
while True:
suffix = getFileSuffix(baseFileName)
if suffix:
suffixes.insert(0, suffix)
baseFileName = baseFileName[:-len('.' + suffix)]
else:
break
return baseFileName + '_%s' % duplicateIdx +\
(('.' +'.'.join(suffixes)) if suffixes else '')
def getGenomeElementSourceClass(fn, suffix=None, forPreProcessor=False):
from gold.application.DataTypes import getUnsupportedFileSuffixes
if not suffix:
from gold.util.CommonFunctions import getFileSuffix
suffix = getFileSuffix(fn)
if not suffix.lower() in getUnsupportedFileSuffixes():
for geSourceCls in getAllGenomeElementSourceClasses(forPreProcessor):
for clsSuffix in geSourceCls.FILE_SUFFIXES:
if (fn.endswith('.' + clsSuffix) if suffix is None else
clsSuffix.lower() == suffix.lower()):
return geSourceCls
else:
if forPreProcessor:
from gold.origdata.GtrackGenomeElementSource import HbGtrackGenomeElementSource
GtrackGESourceCls = HbGtrackGenomeElementSource
else:
from gold.origdata.GtrackGenomeElementSource import GtrackGenomeElementSource
GtrackGESourceCls = GtrackGenomeElementSource
if GtrackGESourceCls.suffixBasedGTrackSubtypeExists(suffix):
return GtrackGESourceCls
raise NotSupportedError('File type "%s" not supported.' % suffix)
def getFileSuffix(self):
return getFileSuffix(self._fn) if self._fn is not None else None
def getTitleSuffix(title):
return getFileSuffix(splitTitleIfDuplicate(title)[0])
def getFileSuffix(self):
return getFileSuffix(self._fn) if self._fn is not None else None
