Source code for src.cmd.util

import os, shutil, MySQLdb, re

from ..segment import all_segments, Segment
from ..configuration import config
__all__ = ['MySQL', 'segment_run_map', 'create_log_dir', 'get_threads', 'validate_project', 'check_pass']

segment_run_map = {'a':'arrest', 
                   'c':'sentence', 
                   'd':'demographic', 
                   's':'supervision',
                   'su':'subject'}

_hasUpper = re.compile('[A-Z]')
_haslower = re.compile('[a-z]')
_hasNum = re.compile('[0-9]')

class MySQL(object):
    def __init__(self, user, pwd):    
        self.conn = MySQLdb.connect(host=config.db_host, user=self.user, passwd=self.pwd, port=config.db_port)
        self.cur = conn.cursor()
    
    def __enter__(self):
        return self

    def __exit__(self, type, value, traceback):
        self.conn.close()
        self.cur.close()

def check_pass(pwd):
    return True if len(pwd) > 6 and _hasUpper.search(pwd) and _hasLower.search(pwd) else False


def create_log_dir(log_dir, replace=False):
    if not os.path.isdir(log_dir): 
        os.makedirs(log_dir)
    elif replace:
        shutil.rmtree(log_dir)
        os.makedirs(log_dir)
    
    return log_dir

def validate_project(project, verbose):
    p_dir = os.path.join(config.root_dir, 'project_data', project)
    
    if not os.path.isdir(p_dir):
        if verbose:
            print 'Invalid project %s' % project
             
        return False
    
    return True

def get_threads(cases, cycles):
    threads = config.transform_threads if config.transform_threads >= 250 else 250 
    """Basic function to determine number of threads to use."""
    _set_min = lambda t: t if t >= 10 else 10
     
    x = cases / 25000 #assume a baseline sample size of 75000
    if x > 3: #adjust if sample size is larger than baseline
        threads = int(threads - ((x - 3)/4)*threads) 
        
    elif x < 3: #for now we do not adjust for smaller sample
        threads = threads 
    
    if _set_min(threads) == 10: return threads #if we set to < 0 return 10 threads
    
    x = cycles / 100 #assume a baseline of 100 cycles
    if x > 1: #adjust if cycles is larger than baseline
        threads = int(threads - ((x - 1)/2)*threads) 
    elif x < 1: #for now we do not adjust for smaller cycles
        threads = threads
    
    print threads
    return _set_min(threads)