def setUp(self): self.hadoop_version = "0.20.2" self.hadoop_version_tuple = (0, 20, 2) self.hadoop_home = tempfile.mkdtemp(prefix="pydoop_test_") self.hadoop_conf = os.path.join(self.hadoop_home, "conf") os.mkdir(self.hadoop_conf) self.bindir = os.path.join(self.hadoop_home, "bin") os.mkdir(self.bindir) self.hadoop_exe = os.path.join(self.bindir, "hadoop") with open(self.hadoop_exe, "w") as fo: fd = fo.fileno() os.fchmod(fd, os.fstat(fd).st_mode | stat.S_IXUSR) fo.write("#!/bin/bash\necho Hadoop %s\n" % self.hadoop_version) self.orig_env = os.environ.copy() self.pf = hu.PathFinder()
Pydoop is a Python interface to Hadoop that allows you to write MapReduce applications and interact with HDFS in pure Python. """ import os import errno import ConfigParser from importlib import import_module import pydoop.hadoop_utils as hu try: from pydoop.config import DEFAULT_HADOOP_HOME from pydoop.version import version as __version__ except ImportError: # should only happen at compile time DEFAULT_HADOOP_HOME = __version__ = None _PATH_FINDER = hu.PathFinder() _HADOOP_INFO = _PATH_FINDER.find() # fill the cache ASAP __author__ = ", ".join(( "Simone Leo", "Gianluigi Zanetti", "Luca Pireddu", "Francesco Cabras", "Mauro Del Rio", "Marco Enrico Piras", )) __author_email__ = ", ".join(( "<*****@*****.**>", "<*****@*****.**>", "<*****@*****.**>", "<*****@*****.**>",