def generateFilterByRegex(regexStrL, convIdxL):
"""
regexStrL :: [str]
Regular expression string list.
convIdxL :: [(str -> ANY), int]
Column index list with converter.
converters will be ignored.
Specify 1 for first element.
0 means all columns.
separator :: str
Separator string of columns in a line.
return :: str -> bool
Filter function.
"""
regexL = map(re.compile, regexStrL)
idxL = map(lambda (_, idx): idx, convIdxL)
project1 = pysows.generateProject(idxL)
def filterByRegex(rec):
"""
rec :: tuple(str)
return :: bool
"""
rec1 = project1(rec)
if len(regexL) < len(rec1):
raise IOError("key length %d > number of regex %d." % (len(rec1), len(regexL)))
predicate = lambda (regex, x): regex.match(x) is not None
return all(map(predicate, zip(regexL, rec1)))
return filterByRegex
def doMain():
args = parseOpts(sys.argv[1:])
outColumnIdxes = map(prefixToIsLeft,
getColumnIndexListWithPrefix(args.out_columns))
getOutRec = generateGetOutputRecord(outColumnIdxes)
lReader = pysows.recordReader(args.left_input, args.separator)
rReader = pysows.recordReader(args.right_input, args.separator)
lKeyIdxL, rKeyIdxL = getKeyIndexLists(args)
lGetKey = pysows.generateProject(lKeyIdxL)
rGetKey = pysows.generateProject(rKeyIdxL)
hashTable = createHashTable(lReader, lGetKey)
resultIter = hashJoin(hashTable, rReader, rGetKey)
for lRec, rRec in resultIter:
pysows.printList(getOutRec(lRec, rRec))
print
