def __init__(self, id, opts, input, output): """Initialize TarsqiControl object conform the data source identifier and the processing options. Does not set the instance variables related to the document model and the meta data. Arguments: id - data source identifier (string) opts - dictionary of processing options input - absolute path output - absolute path""" # Make sure we're in the right directory. If the toolkit # crached on a previous file it may end up being in a # different directory. os.chdir(TTK_ROOT) # initialize options from the settings file self.processing_options = read_settings('settings.txt') # user provided options self.data_source_identifier = id self.processing_options.update(opts) self.input = input self.output = output self.basename = os.path.basename(input) if self.basename.endswith('.xml'): self.basename = self.basename[0:-4] # change the type of some of the processing options _transform_values(self.processing_options) # to be filled in by document model and later processing self.document_model = None self.processing_parameters = None self.metadata = None self.xml_document = None self.document = None # directories and files, user has no control over where # intermediate files go self.DIR_GUTIME = TTK_ROOT + os.sep + 'gutime' self.DIR_DATA = TTK_ROOT + os.sep + 'data' self.DIR_DATA_TMP = self.DIR_DATA + os.sep + 'tmp' self.DIR_PARSED = os.path.join(self.DIR_DATA, 'in', 'parsed') self.FILE_PRE = self.DIR_DATA_TMP + os.sep + self.basename + '.pre.xml' self.FILE_GUT = self.DIR_DATA_TMP + os.sep + self.basename + '.gut.xml' self.FILE_EVI = self.DIR_DATA_TMP + os.sep + self.basename + '.evi.xml' self.FILE_SLI = self.DIR_DATA_TMP + os.sep + self.basename + '.sli.xml' self.FILE_S2T = self.DIR_DATA_TMP + os.sep + self.basename + '.s2t.xml' self.FILE_BLI = self.DIR_DATA_TMP + os.sep + self.basename + '.bli.xml' self.FILE_CLA = self.DIR_DATA_TMP + os.sep + self.basename + '.cla.xml' self.FILE_MER = self.DIR_DATA_TMP + os.sep + self.basename + '.mer.xml' self.collect_parsed_files()
def __init__(self, id, opts, input, output): """Initialize TarsqiControl object conform the data source identifier and the processing options. Does not set the instance variables related to the document model and the meta data. Arguments: id - data source identifier (string) opts - dictionary of processing options input - absolute path output - absolute path""" # Make sure we're in the right directory. If the toolkit # crached on a previous file it may end up being in a # different directory. os.chdir(TTK_ROOT) # initialize options from the settings file self.processing_options = read_settings('settings.txt') # user provided options self.data_source_identifier = id self.processing_options.update(opts) self.input = input self.output = output self.basename = os.path.basename(input) if self.basename.endswith('.xml'): self.basename = self.basename[0:-4] # change the type of some of the processing options _transform_values(self.processing_options) # to be filled in by document model and later processing self.document_model = None self.processing_parameters = None self.metadata = None self.xml_document = None self.document = None # directories and files, user has no control over where # intermediate files go self.DIR_GUTIME = TTK_ROOT + os.sep + 'gutime' self.DIR_DATA = TTK_ROOT + os.sep + 'data' self.DIR_DATA_TMP = self.DIR_DATA + os.sep + 'tmp' self.FILE_PRE = self.DIR_DATA_TMP + os.sep + self.basename + '.pre.xml' self.FILE_GUT = self.DIR_DATA_TMP + os.sep + self.basename + '.gut.xml' self.FILE_EVI = self.DIR_DATA_TMP + os.sep + self.basename + '.evi.xml' self.FILE_SLI = self.DIR_DATA_TMP + os.sep + self.basename + '.sli.xml' self.FILE_S2T = self.DIR_DATA_TMP + os.sep + self.basename + '.s2t.xml' self.FILE_BLI = self.DIR_DATA_TMP + os.sep + self.basename + '.bli.xml' self.FILE_CLA = self.DIR_DATA_TMP + os.sep + self.basename + '.cla.xml' self.FILE_MER = self.DIR_DATA_TMP + os.sep + self.basename + '.mer.xml'
def __init__(self, options): """Initialize options from the settings file and the opts parameter. Loop through the options dictionary and replace some of the strings with other objects: replace 'True' with True, 'False' with False, and strings indicating an integer with that integer.""" self._options = read_settings(SETTINGS) for (option, value) in options: self._options[option[2:]] = value for (attr, value) in self._options.items(): if value in ('True', 'False') or value.isdigit(): self._options[attr] = eval(value) self.genre = self.getopt('genre') self.source = self.getopt('source') self.platform = self.getopt('platform') self.pipeline = self.getopt('pipeline') self.loglevel = self.getopt('loglevel') self.trap_errors = self.getopt('trap-errors', True) self.extension = self.getopt('extension', '') self.perl = self.getopt('perl', 'perl') self.mallet = self.getopt('mallet') self.treetagger = self.getopt('treetagger') self.classifier = self.getopt('classifier') self.ee_model = self.getopt('ee-model') self.et_model = self.getopt('et-model')
if len(sys.argv) < 2: sys.exit("\nMissing platform flag\nUsage: %s\n" % USAGE) args = sys.argv[1:] platform_flag = args.pop(0) try: (platform, ostype) = map(string.strip, platform_flag.split('=')) except ValueError: sys.exit("\nUnexpected argument\nUsage: %s\n" % USAGE) if (platform != 'platform'): sys.exit("\nIllegal first flag.\nUsage: %s\n" % USAGE) if ostype in ('osx', 'linux'): setup_classifier(ostype) else: sys.exit("\nUnsupported os type\nUsage: %s\n" % USAGE) print "Updating settings.txt" flags = {} for arg in args: try: (flag, value) = map(string.strip, arg.split('=')) flags[flag] = value except ValueError: sys.exit("\nUnexpected argument.\nUsage: %s\n" % USAGE) settings = read_settings('settings.txt') add_perl_path(settings) settings.update(flags) write_settings(settings, 'settings.txt')